Saturday, 3 March 2018

Cheap Chinese 70cm 100W Linear Amplifier - Part 1

Wanting to dip my toes deeper into 70cm I decided in December I could do with a little more power than the 50W my Yaesu FT991 puts out and decided a 100W linear was the order of the day.

After looking around on the Internet and various UK suppliers I headed to the drinks cabinet for a stiff whisky after balking at the £400+ price tags and decided a homebrew amp was in order.

A quick Google later and I came across an A/B class linear on eBay, from our friendly cousins in China, based on the Motorola MRF186 for £28.99 inc P&P and promptly ordered the kit, which according the advertisement was 100% brand new.  Two weeks later the kit dropped through the door to say I was impressed is an understatement.

Opening the package up I found a good quality PCB and assorted components however upon closer inspection found the MRF186 was actually second hand along with the High Q capacitors, but otherwise everything looked OK.

70cm kit as supplied
70cm kit as supplied.
Now the fun began as I had never used SMD components before and in fact this was part of the reason for buying such a cheap kit on the basis if I buggered it up I wouldn't have lost much.   Using the much referenced Google again I found the most suggested way to solder the SMD components was to place a blob of solder on one of the pads then using tweezers slide the component into the molten blob then solder the opposite side and hey presto jobs a goodun.

The only major problem was most of the components aren't marked so you need an inductance meter and DVM to determine the sizes.

It took about 45 minutes to complete the kit not including the matching lines, air cored inductors or soldering the MRF in place.

I decided at this point just to check the biasing circuit worked, supplying 24V to the circuit I varied the bias pot and found I got slightly less than 2.5V and not the 5V I was expecting after looking at the schematic.

70cm PA Schematic
70cm Schematic
After much heading scratching and poking around with the voltmeter I found the supposed 5K pot was in fact only 2.5K aggghhh, so it was off to RS for a replacement.

In the mean time I started to wind the required inductors from the supplied 18SWG wire.  The drain chokes went well however the inductors for the LPF didn't, the wire was extremely hard to bend around a 3mm mandrel and in fact I ran out of the supplied wire while trying.  I found some polyurethane coated wire of the same size in my bits box and finally managed to get the LPF coils wound and installed.

Next up was to cut and fix the matching lines which according to the board needed to be 10cm long which were duly cut and soldered to the board.

With the new 5k pot from RS in hand I needed to remove the original 2.5k pot from the board, which without a reflow iron was a pain in the backside.  I found the best way was to use an old pair of side cutters to remove the top of the pot and then split the side of the pot which then meant I could remove the old pot in pieces.   With the new pot soldered in I got a voltage swing of 0 - 4.8V over the full range of the pot.

Next I fixed the PCB to a heatsink and soldered the MRF186 to the board as well as adding a connecting wire so that both drains had access to the power supply.

The completed amplifier
The completed amplifier.

Next I need a 24V 7A power supply so I headed back to eBay and found an LED lighting supply which suited the job and was only £15, be careful with what you order here as some LED lighting supplies are constant current.

According the MRF186 data sheet the idq needs to be set to around 800mA at 900MHz so I set mine to 700mA on the assumption running the MRF186 at 432MHz would possibly result in more gain.

With a dummy load connected to the amp and set my rig to 5W I got 70W out, lucky for me my FT991 has a inbuilt VSWR meter and which showed an VSWR 5.1:1 so I immediately stopped testing, lucky for me both the FT991 an MRF186 can handle a high VSWR for short periods.

The input matching lines looked to be RG316 which with a velocity factor of 0.7 gave me a matching line of 12.2cm at 432MHz, two centimetres longer than the stated 10cm on the PCB so I furnished another line and fitted it and then tested the board again.  The VSWR had fallen to a much more respectable 1.2:1 so I replaced the output matching line as well.

Now when I supplied the 5W of drive I got 80W out and with a quick turn of the gate trimmer brought the output up to 100W, hoorah!

One thing that did cross my mind was the performance of the 7 pole Chebyshev LPF.   The PCB nor the schematic gave a value for the coils so I tested them with my LCR meter, which isn't that accurate below 1mH,  which gave a reading of 20nH.  I modelled the filter in Elsie with a variance of 20% inductance and it looked though the filter would work fine showing the second and third harmonics down 50dB and 70dB respectively.

Low Pass Filter Modelled
LPF modelled in Elsie 20nH inductors

With the amplifier finished I took it an acquaintance of mine who has a Rhode & Schwarz spectrum analyser and measured the spurious emissions which showed that though the third harmonic was -53dBc while the second was only -40dBc with 5W of drive and 100W out.

The UK regulations state that the harmonics in the spurious domain should be at least -70dBc at frequencies above 30MHz, so the amplifier appeared in it's present state not to be suitable for use in the UK.

Once home my friend had sent me an email pointing to an article by Jon Joyce GM4JTJ who had bought the same amplifier.   After reading Jon's article I found he had a similar experience to myself with the matching lines and so I popped him an email seeing if he had also had the same problem with spurious emissions.  Jon sent back a nice email saying he had sold the amplifier awhile ago but from memory had the same results when running at a lower out power of 88W.  He had never got around to cleaning the amp itself up but had used 1/4 wave coax notch filters to remove the unwanted harmonics.

I tested the LPF with my new sdr-kits VNWA.

Yes I know I was trying to save hundreds of pounds by not buying an amplifier only for me to then spend hundreds buying the VNWA but the VNWA is good value and has multiple uses anyway I digress.

The filter looks to be in the right ball park the return loss was horrendous only -12.5dB yielding a VSWR of 1.8:1 or a reflected power of 8.2%.

7 Pole Chebyshev LPF Original Component Values
Original LPF Response

I swapped the capacitors for the first and last poles from 6.8pF to two 3.9pF in parallel and happy days the got a much better return loss of 20.06dB a VSWR of 1.2:1 or a reflected power of 0.8%.

7 Pole Chebyshev LPF Modified First and Last Pole Caps
LPF response with an extra 1pF added to 1st & 7th poles

The data sheet for the MRF186 stated that the device should be run at 28V and not the 24V.  This was a problem as I only had the 24V supply from eBay, but upon closer inspection remembered there was a small pot to adjust the voltage level so with voltmeter in hand I tried the adjustment and low and behold managed to gain the extra 4V.

I measured the 1dB compression point at both voltages and found at 24V and an idq of 700mA the 1dB compression point was 80W for 5W of drive while at 28V and an idq of 800mA it was 110W with 7W of drive so I'll stick with 28V.  The amplifier gives an efficiency of 63% which is about right for a A/B class amplifier.

The next thing to do is test it again on the spectrum analyser but since my friend is on holiday that will need to wait a few weeks.

Part Two coming soon.


  1. Hi there...What kind of heatsink do You use for this module?


    Dubravko 9a5bdp

  2. Just been running one at 170 watts saturated and 50 watts for DATV - really good little amp!